The fat tissue hormone adiponectin (ADP) is predominantly produced by adipocytes and is considered an important link between fat metabolism and glucose regulation (1, 2). ADP regulates the energy homeostasis, through insulin-sensitizing effects (3), upregulation of the expenditure of stored energy and improvement of the lipid metabolism (4), and this seem to affect both the body composition and the glucose metabolism (5). Consistently, ADP was found to be negatively correlated with fat mass and fasting glucose in obese individuals and thus a higher level of ADP is thought to be related to improved glucose metabolism in healthy normal-weight adults (5).

ADP improves insulin secretion and insulin sensitivity (4, 6), meanwhile insulin is a negative regulator of ADP (7, 8). Accordingly, enhanced levels of ADP have been shown in several studies in type 1 diabetes (9, 10). This suggests that increased ADP levels seen in type 1 diabetes is a compensatory mechanism aiming at stimulating beta cell function (9), and a similar compensation may be observed in other types of pancreatogenic diabetes.

Cystic fibrosis (CF) related diabetes (CFRD) is a type of pancreatogenic diabetes characterized by decreased insulin secretion as well as low body mass index (BMI) explained by both low muscle and low fat mass (11). Hence, although type 1 diabetes and CFRD have very different pathophysiology, pancreatic insulin deficiency and hyperglycemia are present in both diseases. Therefore, ADP may also be increased in CFRD, and, in theory, elevated ADP could postpone onset of CFRD by improving both insulin secretion and insulin sensitivity. However, existing data do not fully support this theory. In two studies among adults with CF, Ziai et al. found no clear association between ADP and the total insulin secretion (12) and Hammana et al. reported no association between insulin resistance and ADP among individuals with CF (13). Thus, ADP levels might be influenced by other factors than glucose tolerance in individuals with CF.

We recently reported that individuals with CF and glucose intolerance had reduced BMI (14). Therefore, we hypothesized that the reduction in body mass can confound a potential correlation between ADP and glucose intolerance. This is further supported by Panagopoulou et al., who found that ADP was increased in individuals with a low fat free mass in a study including both individuals with and without CF (11). However, this finding might be biased, as individuals with CF seem to have higher ADP levels compared to non-CF controls (15, 16). Furthermore, no studies have shown that fat mass is negatively correlated with ADP levels in CF.

To our knowledge, existing data in CF populations have related ADP to glucose metabolism and body composition in separate studies. Hence, this supports studying ADP in relation to both glucose metabolism and body composition in the same CF population to expand the current insight in the ADP metabolism in CF. We therefore aimed to assess the association between glucose intolerance, insulin resistance and ADP as well as the associations between measures of body composition (i.e., fat mass, muscle mass, bone mass) and ADP among adults with CF in the Copenhagen CF cohort.

This study on ADP was nested in a cross-sectional study including 134 adults (≥18 years) with CF conducted at the Copenhagen CF center in 2017 prior to the new CF modulator treatment [for details (14)]. The study was approved by the Danish Ethical Committee (H-16022305) and all participants signed the consent after receiving oral and written information about the study. Demographic data, spirometry, and exocrine pancreas function data were obtained from the Danish CF database.

ADP and glucose tolerance were obtained during a standard 2-hour oral glucose tolerance test (OGTT) done during routine visits with clinically stable participants. The participants were fasted for 8 hours before the OGTT. The OGTT included venous blood (i.e., EDTA) samples collected prior, and 1 and 2 hours after intake of 75 g anhydrous glucose. The samples were centrifuged and frozen (-80°C) within 1 hour for the purpose of analyzing all samples in one batch. Fasting plasma (p) was used to assess total ADP levels quantified by duplicate enzyme linked immunosorbent assay analyses (Biovendor, Karasek, Czech Republic). P-insulin (e801 module) and p-glucose (c702 module) were analyzed using a Cobas 8000. Normal glucose tolerance status (NGT), indeterminate glucose tolerance, impaired glucose tolerance (IGT), and CFRD were classified according to the American Diabetes Association CFRD Consensus Statement (17). Homeostatic model assessment for insulin resistance (HOMA-IR) was used as a marker for hepatic insulin resistance (18, 19). Matsuda index equation was used to calculate peripheral insulin sensitivity from the three time points of the OGTT (20, 21), and the 1-hour insulin secretion rate (ISR_1hour) was used as a marker of beta cell function (22). At the day of the OGTT, the C-reactive protein (CRP) (normal range: <10 mg/L) were measured (Cobas 8000, c702 module) as a marker of systemic inflammation.

Body composition was assessed by a dual energy X‐ray absorptiometry (DXA) scanner (Lunar Prodigy Pro, GE Healthcare, Madison, WI, USA). The estimates included total fat mass, muscle mass, and bone mass (bone mineral content). Equivalent to body mass index (BMI, kg/m²), all measurements (mass in kilo) were divided with squared height (m²) to obtain fat-, muscle-, and bone mineral content indices. Weight and height were measured on the day of the DXA scan.

The study included 115 participants, among whom all had fasted blood sampling and a DXA scan conducted and 104 (90%) also consented to an OGTT. Table 1 presents the clinical characteristics with an almost equal number of study participants in each of the four glucose tolerance groups (i.e., NGT, indeterminate glucose tolerance, IGT, and CFRD). The median [IQR] ADP level was 10.0 μg/mL [6.3-12.0 μg/mL].

In models adjusted for age and sex, those with IGT had borderline higher ADP levels compared to NGT (e^β 1.21, 95% CI: 0.98; 1.59, p=0.078) and those with CFRD had higher ADP levels compared to NGT (e^β 1.25, 95% CI: 1.03; 1.53, p=0.026). However, in the fully adjusted models including exocrine pancreas insufficiency, CRP and fat mass index as covariates the estimates for IGT vs NGT (e^β 1.09, 95% CI: 0.85; 1.41) and CFRD vs NGT (e^β 1.14, 95% CI: 0.91; 1.44) were lower. In multivariable analyses, we found no correlation between glucose tolerance groups and ADP ( Figure 1 ). We found no correlation between p-glucose and ADP, while higher levels of p-insulin were correlated with lower levels of ADP, with the strongest correlation seen between fasting p-insulin and ADP ( Table 2 ). Further, insulin sensitivity (inverse HOMA-IR and Matsuda index) was correlated with higher levels of ADP, while we found no correlation between insulin secretion (ISR_1hour) and ADP levels ( Table 2 ).

BMI, fat mass index, muscle mass index, and bone mineral content index were all negatively correlated with ADP in both minimal adjusted analyses and multivariable analyses indicating that for each one kg/m² increase in BMI, fat mass index, and muscle mass index the levels of ADP decreased by 4-9% (p<0.05, Table 3 ). In multivariable sensitivity analyses including both fat mass index and muscle mass index, the negative correlation between fat mass index and ADP persisted although it was no longer statistically significant; ADP was 3% lower (e^β 0.97/kg/m², 95% CI: 0.94; 1.00, p=0.078) for each one unit (kg/m²) increase in fat mass index. The correlations between muscle mass index and ADP (e^β 0.93/kg/m², 95% CI: 0.88; 0.97, p<0.001) and between bone mineral content index and ADP (e^β 0.23/kg/m², 95% CI: 0.10; 0.53, p<0.001) were not affected by adjustment with fat mass index.

In this study among adults with CF, we found that ADP levels were unrelated to p-glucose and insulin secretion rate, while ADP was inversely correlated with p-insulin levels and positively correlated with insulin sensitivity. Moreover, ADP was negatively correlated with BMI, fat mass, muscle mass, and bone mass.

Overall, severe glucose intolerance was associated with higher ADP levels, although this was partially confounded by pancreas insufficiency, inflammation, and fat mass index. Hence, the association was likely explained by fat mass rather than glucose intolerance. This is similar to previous studies in CF showing that ADP did not change with glucose tolerance status (12, 13). Likewise, we did not find any correlation between insulin secretion rate and ADP, which is also comparable to a former study in individuals with CF (12). However, we did see that the measured p-insulin levels and markers of insulin sensitivity were correlated with ADP and not affected by adjustment with fat mass. HOMA-IR, usually considered a marker of hepatic insulin resistance (23), seemed to be stronger linked to levels of ADP compared to the Matsuda index, which is a better marker of peripheral insulin sensitivity (24). Yet, CF liver disease does not seem to explain the inverse correlation between hepatic insulin resistance and ADP, as CF liver disease has been linked to elevated levels of ADP in CF (11). In other words, ADP is unlikely to explain the link between CF liver disease and glucose intolerance (25) as well as the relationship between reduced insulin sensitivity and CFRD (26). This indicates that CFRD should be differentiated from non-alcoholic fatty liver disease and type 2 diabetes and treatments targeting these diseases should therefore not be used as CFRD management in lean individuals without thorough evaluation.

We also found that all measures and compartments of body mass were negatively correlated with ADP, which is in line with a recent CF study from Qatar (27), but in contrast with a previous CF study from Canada, which reported no correlation between BMI and ADP (13). However, it is not clear from the previous study whether the model was adjusted for sex and age, which is important, since cofactors such as female sex (13) and increasing age (28) are correlated with increased levels of ADP. Further, we showed that fat mass index was negatively correlated with ADP, which seems to be in contrast to a study by Panagopoulou et al., who found a positive association between body fat percentage and ADP (11). However, their study included non-CF controls, thus our study design is not comparable to their study design and differences should be interpreted with caution. Although our findings deviated from previous studies in CF, we found similar trends as is known from several non-CF studies reporting negative associations between fat mass and ADP (29–31). Furthermore, we found increased muscle mass to be correlated with low ADP, unrelated to the fat mass. This is in line with data from Polito et al., who demonstrated that frequent physical exercise lead to lower levels of ADP (15). Similarly, in a non-CF study among individuals with a high prevalence of sarcopenia, both muscle mass index and muscle strength were negatively correlated with ADP (32). This highlights that though ADP levels are strongly connected to body composition and most likely plays important regulatory roles in CF metabolism, ADP appears to be less valuable for evaluating nutritional status in clinical settings.

Following the introduction of new highly effective modulator therapy in the CF population, real-life studies have found significant increases in body mass (33). This change is likely to lead to a reduction in ADP, potentially worsening hepatic insulin sensitivity. Although, hepatic insulin sensitivity does not appear to be the cause of CFRD, it could further complicate disease severity and management. This suggests that incretin-based therapies may play a role in CFRD treatment in near future, as this class of drug seems to enhance ADP levels during treatment (34). However, such interventions should be watchfully introduced, as the current literature provides little knowledge about what role adiponectin plays in pulmonary inflammation and lung function deterioration, and increasing ADP may therefore have both protective (35) or harmful effects on lung function (27).

Finally, we did see that lower bone mass was correlated with higher ADP levels. Although CF bone disease and osteoporosis are common and the results mimic data from individuals with anorexia nervosa, where osteoporosis was linked to higher ADP levels (36), these data should be interpreted with caution. If any true association do exist between bone mass and ADP, we speculate that ADP could stimulate osteoclast formation and thus reduce bone mass, as has been seen in in vitro studies (37). Even so, as individuals with CF are often malnourished, reduced fat and muscle mass might lead to increased ADP as well as a subsequent loss of bone mass and thus correlations between bone and ADP might be a chance finding due to the impact from other body compartments, which strongly correlate with bone mass.

As ADP may have a role as an anti-inflammatory agent (3), chronic inflammation could have been correlated with lower levels of ADP in CF (16). However, we found no impact from inflammation in our study as adjusting for CRP did not affect the correlations between body mass and ADP, and the CRP was generally low, and participants were clinically stable without signs of pulmonary exacerbations. At the same time, a limitation in this study is that we only measured total ADP without assessing the high molecular weight isoform, a sub-component of ADP suggested to be a superior marker of insulin sensitivity (38). Moreover, despite the adjustment for fat mass, the results could still be partly confounded by the effect of fat mass, as fasting insulin, HOMA-IR, muscle, and bone mass are all positively correlated with fat mass and difficult to separate completely. Lastly, as this was a cross-sectional study, all mentioning of causal relations are speculative.

Among adults with CF, we found no correlation between glucose tolerance and ADP, although insulin sensitivity was correlated with higher levels of ADP. Moreover, we found that all measures of body composition were negatively correlated with ADP similar to non-CF populations. This suggests that malnutrition irrespective of CF is associated with higher ADP levels and with increasing body weight after the new highly effective modulator therapy, factors such as fat mass, ADP levels and hepatic insulin sensitivity may become more important in the development of glucose intolerance in CF.